By Gokul Siddharthan J, DCMME Graduate Student Assistant


There have been many promises by various companies that driverless cars will be ploughing through our streets by the end of this decade. Yet, the progress that has been made until now suggests otherwise. Waymo, the current industry leader in driverless cars, promised by the end of 2018 a driverless-taxi service in Phoenix, Arizona. The plan has been a damp squib. Only a few customers, who have signed non-disclosure agreements, and a few parts of the city have been covered.

Driverless cars have been the flagship application for the power of AI. The progress made in the power of technology has not yet solved the ability to mimic human intelligence. Machine learning is an important component of driverless technologies. Millions of miles of practice have still not taught computers to overcome what is called edge cases. Edge cases are unique situations that happen on the roads, such as a light aircraft landing on a busy road. Though situations may not be that dramatic, such cases cannot be avoided and a driverless car would need to think on its feet to maneuver around. Humans need far less practice. Couple lessons and few hundred miles of practice is enough for anyone to be able to drive around the world, be it a busy road in New Delhi or a gravel track in rural Greece. Driverless technology has many milestones to reach. In September, Morgan Stanley cut the valuation of Waymo by 40%, to $105bn, citing delays in its technology. Some firms in China have even suggested redesigning entire cities for driverless cars is far easier.

Most technologies, what is currently called AI, are both powerful and limited. Progress has been transformative, at the same time, the eventual goal of human-like intelligence seems distant. People need to reduce their expectations and not fall for the usual hyperbole. There is little doubt that driverless cars will be in the future, but the consensus is it’s not imminent.